When should I use semantic versus episodic storage?

Use Semantic storage for permanent facts like project requirements or tech stacks, and Episodic storage for chronological events like 'Fixed the login bug' or 'Discussed API changes'.

Can I query the memory system using SQL?

Yes, the skill provides a CLI helper and direct database access for running raw SQL queries and specialized functions for memory retrieval and management.

What types of memory does this skill support?

It supports five distinct types: Semantic (facts and knowledge), Episodic (events and history), Procedural (how-to steps), Strategic (patterns), and Working memory (temporary session context).

How is the data actually stored?

The system typically runs in a Docker environment using a PostgreSQL database with the Apache AGE extension to handle complex memory relationships and embeddings.

How does context packing work?

The system uses token-aware retrieval to pack the most relevant memories into a markdown format, ensuring you provide the LLM with maximum context without exceeding token limits.

AGI Memory System

Name: AGI Memory System
Author: bigadamknight

bybigadamknight

데이터베이스 관리

Manages persistent AI memory through semantic, episodic, and working storage to enhance long-term context and knowledge recall.

소개

The AGI Memory System skill provides a robust framework for managing AI cognition and long-term knowledge retention within Claude Code. It enables users to store and retrieve various tiers of memory—including semantic facts, episodic events, and temporary working context—using dedicated MCP tools or a CLI helper. By leveraging token-aware context packing and semantic search, this skill ensures that Claude maintains a high-fidelity understanding of user preferences, project history, and procedural strategies across multiple sessions, effectively solving the problem of context loss in long-term development projects.

주요 기능

0 GitHub stars
Token-aware context packing to optimize LLM prompt efficiency
Direct MCP tool integration for seamless memory recall and storage
Multi-tier memory architecture (Semantic, Episodic, Strategic, and Working)
Semantic search capabilities with confidence and importance weighting
Built-in health monitoring and database management utilities

사용 사례

Managing short-term task context that automatically expires to keep prompts focused
Tracking a detailed history of debugging attempts and architectural decisions
Persisting user-specific coding preferences and project standards across separate sessions

소개

주요 기능

0 GitHub stars
Token-aware context packing to optimize LLM prompt efficiency
Direct MCP tool integration for seamless memory recall and storage
Multi-tier memory architecture (Semantic, Episodic, Strategic, and Working)
Semantic search capabilities with confidence and importance weighting
Built-in health monitoring and database management utilities

사용 사례

Managing short-term task context that automatically expires to keep prompts focused
Tracking a detailed history of debugging attempts and architectural decisions
Persisting user-specific coding preferences and project standards across separate sessions